DMAR: [INTR-REMAP] Blocked an interrupt request due to source-id verification failure

Submitted by Nikolay Kichukov

Assigned to Default DRI bug account

Link to original bug (#111455)

Description

Hello team,
the below error is printed in the log and causes the graphics driver to freeze:

[Tue Aug 20 12:04:38 2019] DMAR: DRHD: handling fault status reg 2
[Tue Aug 20 12:04:38 2019] DMAR: [INTR-REMAP] Request device [00:00.0] fault index 26 [fault reason 38] Blocked an interrupt request due to source-id verification failure
[Tue Aug 20 12:04:38 2019] [drm] Fence fallback timer expired on ring gfx
[Tue Aug 20 12:04:38 2019] [drm] Fence fallback timer expired on ring gfx
[Tue Aug 20 12:04:39 2019] [drm] Fence fallback timer expired on ring gfx
[Tue Aug 20 12:04:39 2019] [drm] Fence fallback timer expired on ring gfx
[Tue Aug 20 12:04:40 2019] [drm] Fence fallback timer expired on ring gfx
[Tue Aug 20 12:04:40 2019] [drm] Fence fallback timer expired on ring sdma0
[Tue Aug 20 12:04:41 2019] [drm] Fence fallback timer expired on ring sdma0
...

Hardware is: Dell Precision Tower 5810 with Advanced Micro Devices, Inc. [AMD/ATI] Oland GL [FirePro W2100] video card.

Kernel: 5.2.8 x86_64 (GNU/Gentoo Linux)
Kernel Command line: BOOT_IMAGE=/kernel-genkernel-x86_64-5.2.8-gentoo root=/dev/mapper/root ro crypt_root=UUID=e11887f5-4104-4a9e-9c53-7e1d904a0b28 root_trim=no elevator=bfq scsi_mod.use_blk_mq=1 libata.allow_tpm=1 domdadm dolvm intel_iommu=on

IOMMU is on, because the system acts mainly as a KVM/Libvirt host.

amdgpu driver information:

dmesg | grep amd

[ 14.614307] [drm] amdgpu kernel modesetting enabled.
[ 14.615206] amdgpu 0000:03:00.0: remove_conflicting_pci_framebuffers: bar 0: 0xe0000000 -> 0xefffffff
[ 14.615208] amdgpu 0000:03:00.0: remove_conflicting_pci_framebuffers: bar 2: 0xf7e00000 -> 0xf7e3ffff
[ 14.615209] amdgpu 0000:03:00.0: vgaarb: deactivate vga console
[ 14.617260] amdgpu 0000:03:00.0: kfd not supported on this ASIC
[ 14.622071] amdgpu 0000:03:00.0: No more image in the PCI ROM
[ 14.624585] amdgpu 0000:03:00.0: VRAM: 2048M 0x000000F400000000 - 0x000000F47FFFFFFF (2048M used)
[ 14.624586] amdgpu 0000:03:00.0: GART: 1024M 0x000000FF00000000 - 0x000000FF3FFFFFFF
[ 14.626166] [drm] amdgpu: 2048M of VRAM memory ready
[ 14.626169] [drm] amdgpu: 3072M of GTT memory ready.
[ 14.626832] amdgpu 0000:03:00.0: PCIE GART of 1024M enabled (table at 0x000000F400900000).
[ 14.643922] [drm] amdgpu: dpm initialized
[ 14.862039] fbcon: amdgpudrmfb (fb0) is primary device
[ 14.990216] amdgpu 0000:03:00.0: fb0: amdgpudrmfb frame buffer device
[ 15.260648] [drm] Initialized amdgpu 3.32.0 20150101 for 0000:03:00.0 on minor 0

and module dependencies:

lsmod | grep amdgpu

amdgpu 3772416 7
gpu_sched 36864 1 amdgpu
ttm 114688 1 amdgpu
drm_kms_helper 212992 1 amdgpu
drm 462848 7 gpu_sched,drm_kms_helper,amdgpu,ttm
i2c_algo_bit 16384 2 igb,amdgpu

Happy to collect output from 'drm.debug=0x1e log_buf_len=4M' booted kernel if that makes sense.

Thanks,
-Nikolay